Record locking

[Bill Vermillion]

On Wed, Mar 31, 2004 at 01:24:38PM -0600, Mike Schwartz-PC
Support & Services thus spoke:

> > In other words, regarding programming code, there is (or
> > should never be) any such thing as "timing problems". In
> > fact, I don't know what "timing problems" means in the way
> > you're using it.

>  Well, when the hardware doesn't respond in the way the
> software expects it to, I guess you could call that a "timimg
> problem."

>  For example, on some of the Xenix versions when we used to
> do floppy disk backups, the software was prompting the users
> to insert the next disk, but users were pulling the diskettes
> out before the disk writes had completed. I would call this a
> hardware/software timing problem.

That was failure of OS design and commodity hardware.  Good code
and hardware would have locked the floppy in place.  You will see
this in current things with CD ROMs or tape drives.  The original
8" floppies and I think at least some of the 5.25" devices did
this.   But to make things cheaper safe guards were taken out.

You're earlier post on not being able to run Xenix on high-speed
machines was because of inherent timing loops.  When Xenix was
developed not everything had RTCs so they only way to get any
control over anything was to run loops on the CPU clocks.

Up until recently there were problems on installs of the OSR5 on
the Pentium IV chips set because of some timing loops that were
causing CPU heating.   A P4 will throttle itself down to a slower
clock speed when it starts heating in order to prevent thermal
runaway.  Jeff Lieberman said he first noticed this on one machine
that at each reboot during some work the system would boot up at a
lower clock speed, eventually getting down to about 500MHz on
a 2GHz+ machine. 

I recently had some quirkiness on a P4 XP and to be safe I
rebooted, and the the CPUs detected [1 CPU with two cores in HTT
mode] both showed 1.2GHz speed and not the 2.4GHz speed.  I checked
and something fell in front of the case fans, leaving the system
running with only the PS fan and the CPU fan.

>  The other two things problems we've been discussing (records
> still showing up as being locked to other users even though we
> have issued a WRITE command and the inability to run multiple
> *report lines in a Windows batch file) are two such hardware
> timing errors that we have simply learned to program around.

And what versions of Windows.  The newer versions are certainly
more stable and network aware than those of the past.

>  If I thought there was any reasonable way that fPtech could
> program around such hardware speed limitations on a few
> specific computers or on a few versions of Windows, I would
> certainly submit all this stuff as a bug, even if I had to rig
> up a machine that demonstrates the problem and send it to him
> to play with...

And what versions of Windows would this be.  There are many of
them.  As to specific computers - you can only be sure you are
getting close the same HW when you go with the largest name brands
- eg IBM, Compaq, HP, as other brands - the Dells of the world and
their counterparts will change HW the moment they get a better
price.

>  Then again, we all have better things to do with our time
> than to try and troubleshoot problems that we can easily work
> around, especially where answers have been posted to this list
> on these problems several times.

The easy work arounds aren't a problem :-)

Bill
-- 
Bill Vermillion - bv @ wjv . com